- Email: [email protected]
PREDICTORS OF POST-OPERATIVE DELIRIUM IN CARDIAC SURGERY PATIENTS; A MACHINE LEARNING APPROACH
Abstracts
S237
upon their lived experiences during a realistic but riskfree environment. Additionally, these exercises serve as an ice-breaker that helps dissolve preconceived notions and expectations of interprofessional relationships. The benefits of offering this type of training with simulated exercises to existing surgical teams may also realize similar improvements.
357 SECONDARY PREVENTION AFTER CORONARY ARTERY BYPASS GRAFT SURGERY: PRESENTATION OF A SCIENTIFIC STATEMENT M Ruel, A Kulik Ottawa, Ontario BACKGROUND:
A proven therapy for nearly 50 years, CABG is the most durable and complete treatment of ischemic heart disease. However, in the months and years that follow surgery, patients who have undergone CABG remain at risk for subsequent ischemic events as a result of native coronary artery disease progression and the development of graft atherosclerosis. Secondary therapies therefore play a key role in the maintenance of native and graft vessel patency and for the prevention of adverse cardiovascular outcomes. METHODS AND RESULTS: We present a Scientific Statement prepared for the American Heart Association (AHA) to expand on two 2011 AHA and American College of Cardiology documents that provided a general overview of secondary prevention and briefly summarized the use of medical therapy after surgical coronary revascularization. Since the writing of those two statements, important evidence from clinical and observational trials has emerged that further supports and broadens the merits of intensive risk-reduction therapies for CABG patients. The present statement, specifically focused on the CABG population, thoroughly evaluates the current state of evidence regarding preventative therapies after surgery. Postoperative antiplatelet agents and lipid-lowering therapy continue to be the mainstay of secondary prevention following coronary surgical revascularization. Other opportunities for improving long-term clinical outcomes after CABG include the aggressive management of hypertension and diabetes mellitus, smoking cessation, weight loss, and cardiac rehabilitation. Select Class I and IIa recommendations are summarized in the Table below. CONCLUSION: In addition to providing specific, revised and updated recommendations regarding the use of secondary preventive therapies following CABG, this statement will also highlight areas in need of prospectively collected clinical data.
358 PREDICTORS OF POST-OPERATIVE DELIRIUM IN CARDIAC SURGERY PATIENTS; A MACHINE LEARNING APPROACH HN Mufti, S Abidi, SR Abidi, GM Hirsch Halifax, Nova Scotia BACKGROUND:
Delirium occurs relatively frequently (10 to 15%) among patients who undergo cardiac surgery. Patients who experience delirium are at higher risk of other adverse outcomes. Identification of patients at risk will allow us to target preventive strategies and improve informed consent. The purpose of this study was to compare two novel Machine Learning (ML) data analysis approach, namely Artificial Neural Networks (ANN) and Bayesian Networks (BN), against standard Logistic Regression modeling (LogR) in predicting delirium among cardiac surgery patients.
S238 METHODS:
The Maritime Heart Center (MHC) registry is a comprehensive database covering all cardiac surgery patients. We examined data of patients who underwent CABG and or valve surgery between 2006 and 2012. Delirium was defined as a temporary mental disturbance requiring medical or physical intervention. Dataset were split randomly into Derivation (80%) and validation (20%) subsets. In this analysis, we used two ML statistical tools: ANN, which is a non-linear technique that models complex relationships, and BN, which is a type of graphical models that provide a natural tool for dealing with uncertainty and complexity. These results were compared with a conventional stepwise LogR. RESULTS: Among the 5,584 patients who met the study criteria, 634 (11.4%) experienced delirium. Interestingly, both approaches produced similar predictive variables for delirium (e.g., age, CVD & blood transfusion), although the variable weighting was different. Using the validation set, specificity was excellent across all models (95.8% for ANN, 93.3% for BN and 97.7% for LogR). However, the positive predictive value was 46.1% for ANN, 41.5% for BN and 33.3% for LogR. Sensitivity was 27.5% for ANN, 37% for BN and only 8.6% for LogR. Receiver operator curves were 76.6% for ANN, 76.4% for BN and 77.2% for LogR. CONCLUSION: In this study, we successfully demonstrate the use of statistical machine learning Methods (namely: ANN and BN) in the prediction of delirium. ML Methods generated superior results compared to standard LogR. ML techniques offer novel and alternative approaches to aid in the prediction of complex relationships, which is a typical feature of adverse medical events.
359 SUTURELESS AORTIC VALVE REPLACEMENT USING THE PERCEVAL S PROSTHESIS: SHOULD BICUSPID DISEASE BE A CONTRAINDICATION?
Canadian Journal of Cardiology Volume 30 2014
were collected from preoperative and postoperative transthoracic echocardiography reports (maximum 28 days postop) and from intensive care unit charts. Surgical approach consisted in a median sternotomy in 13 patients (52%), partial sternotomy in 5 (20%) and minithoracotomy in 7 (28%). Due to advanced age and presence of comorbidities, these patients were selected for AVR. RESULTS: Sutureless AVR was successfully performed in all patients (n¼25, 100%). The study population included 17 (68%) male and 8 (32%) female, with a median age of 77.8 5.4 years old. Eleven (44%) of these patients had a preoperative New York Heart Association functional class of III-IV/ IV. The mean Euroscore II was 3.4 2.6 %. Concomitant procedures included CABG in 8 patients (32%), ascending aortic replacement in 2 (8%), mitral valve repair in 1 (4%), septal myomectomy in 1 (4%) and atrial septal defect closure in 1 (4%). Mean aortic cross-clamp time was 46 14 minutes in patients who had sutureless AVR alone and 56 14 minutes in those who had concomitant procedures. Mean trans-aortic valve gradient decreased from an average of 49.4 15.7 mmHg preoperatively to 14.5 5.4 mmHg postoperatively. Mean aortic valve area increased from 0.78 0.18 cm2 preoperatively to 1.75 0.43 cm2 postoperatively. Five patients suffered from atrio-ventricular block requiring permanent pacemaker implantation. Two patients suffered from stroke, one of which resulted in permanent hemiparesis. There was no paravalvular leak at the pre-discharge TTE. No postoperative valve migration or embolization was reported. In-hospital mortality occurred in one patient (4%). Mean ICU length of stay was 3 2 days postoperatively. CONCLUSION: This study demonstrates that Perceval S sutureless valve can be deployed in patients with bicuspid aortic disease without increasing the risk of paravalvular leak, allowing short cross-clamp times, satisfactory postoperative mean trans-aortic gradients and good perioperative mortality and morbidity, considering the high baseline risk status of this population. Bicuspid aortic valve disease should not be considered a contraindication to sutureless AVR.
360 PREOPERATIVE CREATININE CLEARANCE AFFECTS LONG-TERM SURVIVAL AFTER OFF-PUMP CORONARY ARTERY BYPASS SURGERY N Vistarini, A Deschamps, R Cartier
A Nguyen, W Fortin, A Mazine, D Bouchard, P Demers
Montréal, Québec
Montréal, Québec
BACKGROUND:
BACKGROUND:
Sutureless aortic valve replacement (AVR) is a relatively new technique for the treatment of aortic stenosis. Bicuspid aortic valve is generally considered a contraindication to sutureless AVR. The aim of this study was to evaluate the feasibility and perioperative outcomes of this technique in patients with bicuspid aortic disease. METHODS: Between June 2011 and January 2014, 25 patients with aortic bicuspid disease underwent sutureless AVR. Data
Chronic renal failure (CRF) is a major risk factor for long-term survival after coronary surgery. The objective of this study is to assess the influence of the severity of preoperative CRF on survival after off-pump coronary artery bypass (OPCAB) surgery. METHODS: Data on 1400 consecutive patients who underwent off-pump isolated CABG between September 1996 and April 2007 were prospectively collected. Preoperative creatinine clearance (CrCl) was estimated according to Cockcroft and
S237
upon their lived experiences during a realistic but riskfree environment. Additionally, these exercises serve as an ice-breaker that helps dissolve preconceived notions and expectations of interprofessional relationships. The benefits of offering this type of training with simulated exercises to existing surgical teams may also realize similar improvements.
357 SECONDARY PREVENTION AFTER CORONARY ARTERY BYPASS GRAFT SURGERY: PRESENTATION OF A SCIENTIFIC STATEMENT M Ruel, A Kulik Ottawa, Ontario BACKGROUND:
A proven therapy for nearly 50 years, CABG is the most durable and complete treatment of ischemic heart disease. However, in the months and years that follow surgery, patients who have undergone CABG remain at risk for subsequent ischemic events as a result of native coronary artery disease progression and the development of graft atherosclerosis. Secondary therapies therefore play a key role in the maintenance of native and graft vessel patency and for the prevention of adverse cardiovascular outcomes. METHODS AND RESULTS: We present a Scientific Statement prepared for the American Heart Association (AHA) to expand on two 2011 AHA and American College of Cardiology documents that provided a general overview of secondary prevention and briefly summarized the use of medical therapy after surgical coronary revascularization. Since the writing of those two statements, important evidence from clinical and observational trials has emerged that further supports and broadens the merits of intensive risk-reduction therapies for CABG patients. The present statement, specifically focused on the CABG population, thoroughly evaluates the current state of evidence regarding preventative therapies after surgery. Postoperative antiplatelet agents and lipid-lowering therapy continue to be the mainstay of secondary prevention following coronary surgical revascularization. Other opportunities for improving long-term clinical outcomes after CABG include the aggressive management of hypertension and diabetes mellitus, smoking cessation, weight loss, and cardiac rehabilitation. Select Class I and IIa recommendations are summarized in the Table below. CONCLUSION: In addition to providing specific, revised and updated recommendations regarding the use of secondary preventive therapies following CABG, this statement will also highlight areas in need of prospectively collected clinical data.
358 PREDICTORS OF POST-OPERATIVE DELIRIUM IN CARDIAC SURGERY PATIENTS; A MACHINE LEARNING APPROACH HN Mufti, S Abidi, SR Abidi, GM Hirsch Halifax, Nova Scotia BACKGROUND:
Delirium occurs relatively frequently (10 to 15%) among patients who undergo cardiac surgery. Patients who experience delirium are at higher risk of other adverse outcomes. Identification of patients at risk will allow us to target preventive strategies and improve informed consent. The purpose of this study was to compare two novel Machine Learning (ML) data analysis approach, namely Artificial Neural Networks (ANN) and Bayesian Networks (BN), against standard Logistic Regression modeling (LogR) in predicting delirium among cardiac surgery patients.
S238 METHODS:
The Maritime Heart Center (MHC) registry is a comprehensive database covering all cardiac surgery patients. We examined data of patients who underwent CABG and or valve surgery between 2006 and 2012. Delirium was defined as a temporary mental disturbance requiring medical or physical intervention. Dataset were split randomly into Derivation (80%) and validation (20%) subsets. In this analysis, we used two ML statistical tools: ANN, which is a non-linear technique that models complex relationships, and BN, which is a type of graphical models that provide a natural tool for dealing with uncertainty and complexity. These results were compared with a conventional stepwise LogR. RESULTS: Among the 5,584 patients who met the study criteria, 634 (11.4%) experienced delirium. Interestingly, both approaches produced similar predictive variables for delirium (e.g., age, CVD & blood transfusion), although the variable weighting was different. Using the validation set, specificity was excellent across all models (95.8% for ANN, 93.3% for BN and 97.7% for LogR). However, the positive predictive value was 46.1% for ANN, 41.5% for BN and 33.3% for LogR. Sensitivity was 27.5% for ANN, 37% for BN and only 8.6% for LogR. Receiver operator curves were 76.6% for ANN, 76.4% for BN and 77.2% for LogR. CONCLUSION: In this study, we successfully demonstrate the use of statistical machine learning Methods (namely: ANN and BN) in the prediction of delirium. ML Methods generated superior results compared to standard LogR. ML techniques offer novel and alternative approaches to aid in the prediction of complex relationships, which is a typical feature of adverse medical events.
359 SUTURELESS AORTIC VALVE REPLACEMENT USING THE PERCEVAL S PROSTHESIS: SHOULD BICUSPID DISEASE BE A CONTRAINDICATION?
Canadian Journal of Cardiology Volume 30 2014
were collected from preoperative and postoperative transthoracic echocardiography reports (maximum 28 days postop) and from intensive care unit charts. Surgical approach consisted in a median sternotomy in 13 patients (52%), partial sternotomy in 5 (20%) and minithoracotomy in 7 (28%). Due to advanced age and presence of comorbidities, these patients were selected for AVR. RESULTS: Sutureless AVR was successfully performed in all patients (n¼25, 100%). The study population included 17 (68%) male and 8 (32%) female, with a median age of 77.8 5.4 years old. Eleven (44%) of these patients had a preoperative New York Heart Association functional class of III-IV/ IV. The mean Euroscore II was 3.4 2.6 %. Concomitant procedures included CABG in 8 patients (32%), ascending aortic replacement in 2 (8%), mitral valve repair in 1 (4%), septal myomectomy in 1 (4%) and atrial septal defect closure in 1 (4%). Mean aortic cross-clamp time was 46 14 minutes in patients who had sutureless AVR alone and 56 14 minutes in those who had concomitant procedures. Mean trans-aortic valve gradient decreased from an average of 49.4 15.7 mmHg preoperatively to 14.5 5.4 mmHg postoperatively. Mean aortic valve area increased from 0.78 0.18 cm2 preoperatively to 1.75 0.43 cm2 postoperatively. Five patients suffered from atrio-ventricular block requiring permanent pacemaker implantation. Two patients suffered from stroke, one of which resulted in permanent hemiparesis. There was no paravalvular leak at the pre-discharge TTE. No postoperative valve migration or embolization was reported. In-hospital mortality occurred in one patient (4%). Mean ICU length of stay was 3 2 days postoperatively. CONCLUSION: This study demonstrates that Perceval S sutureless valve can be deployed in patients with bicuspid aortic disease without increasing the risk of paravalvular leak, allowing short cross-clamp times, satisfactory postoperative mean trans-aortic gradients and good perioperative mortality and morbidity, considering the high baseline risk status of this population. Bicuspid aortic valve disease should not be considered a contraindication to sutureless AVR.
360 PREOPERATIVE CREATININE CLEARANCE AFFECTS LONG-TERM SURVIVAL AFTER OFF-PUMP CORONARY ARTERY BYPASS SURGERY N Vistarini, A Deschamps, R Cartier
A Nguyen, W Fortin, A Mazine, D Bouchard, P Demers
Montréal, Québec
Montréal, Québec
BACKGROUND:
BACKGROUND:
Sutureless aortic valve replacement (AVR) is a relatively new technique for the treatment of aortic stenosis. Bicuspid aortic valve is generally considered a contraindication to sutureless AVR. The aim of this study was to evaluate the feasibility and perioperative outcomes of this technique in patients with bicuspid aortic disease. METHODS: Between June 2011 and January 2014, 25 patients with aortic bicuspid disease underwent sutureless AVR. Data
Chronic renal failure (CRF) is a major risk factor for long-term survival after coronary surgery. The objective of this study is to assess the influence of the severity of preoperative CRF on survival after off-pump coronary artery bypass (OPCAB) surgery. METHODS: Data on 1400 consecutive patients who underwent off-pump isolated CABG between September 1996 and April 2007 were prospectively collected. Preoperative creatinine clearance (CrCl) was estimated according to Cockcroft and